# Qwen4-Max

Qwen4-Max is Alibaba's latest language model in our dataset, released Nov 16, 2026. Proprietary, API access only. Alibaba's flagship Qwen4 multimodal model, released Nov 16, 2026: a 2.4-trillion-parameter mixture-of-experts design that activates 71 billion parameters per token, with a one-million-token context window, native image and video understanding, and QwenCloud list prices of $1.50 input, $4.50 output and $0.075 implicitly cached input per 1M tokens. Verified against vendor sources on Nov 16, 2026.

Primary source (source of record; Agentic Weekly is the fast independent index): https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783

Last verified: 2026-11-16 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/qwen4-max · JSON: https://theagenticweekly.com/api/models/qwen4-max

## Key facts

| Fact | Value | Source | Label | Anchor |
|---|---|---|---|---|
| Vendor | Alibaba |  | editorial | https://theagenticweekly.com/models/qwen4-max#fact-vendor |
| Released | Nov 16, 2026 | https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783 | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-release-date |
| API id | `qwen4-max` | https://www.qwencloud.com/models/qwen4-max | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-api-id |
| Context window | 1,000,000 tokens (1M). Stated as 1M on the QwenCloud model page and as a one-million-token context window in the Alibaba Cloud Community announcement. | https://www.qwencloud.com/models/qwen4-max | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-context-tokens |
| Max output | 393,000 tokens. Stated as 393K for both standard and thinking output on the QwenCloud model page. Max input is stated as 991K, or 983K in thinking mode, with a 262K max reasoning budget. | https://www.qwencloud.com/models/qwen4-max | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-max-output-tokens |
| Input | text, image, video | https://www.qwencloud.com/models/qwen4-max | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-input-modalities |
| Output | text | https://www.qwencloud.com/models/qwen4-max | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-output-modalities |
| Parameters | 2.4T total, 71B active, mixture-of-experts | https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783 | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-parameters |
| Licence | Proprietary (API). Available on QwenCloud via the Token Plan and API, and on Alibaba platforms including QwenWork and QwenStudio. Open-weight smaller Qwen4 models are described as forthcoming. | https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783 | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-license |
| Pricing | $1.5 in / $4.5 out per 1M tokens. QwenCloud list prices. Cached input is the implicit-cache rate. The same page also lists explicit cache creation and read rates, which are not in these three fields. | https://www.qwencloud.com/models/qwen4-max | vendor-reported, confirmed | https://theagenticweekly.com/models/qwen4-max#fact-pricing |

## Vendor claims and neutral signals

- Vendor claim (Alibaba): ranked 6th on Code Arena, a real-world coding leaderboard. We did not verify this.
- Vendor claim (Alibaba): on Work360, a proprietary benchmark across 721 specialized fields, Qwen4-Max outperforms Qwen3.8-Max overall while using 35% fewer tokens at High effort reasoning mode and reducing the average cost per task by around 60%. We did not verify this.

## What we tested

Nothing yet. We have not run Qwen4-Max ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- The QwenCloud model page also lists 5-minute explicit cache creation at $1.875 per 1M tokens, 1-hour explicit cache creation at $3 per 1M tokens, and explicit cache read at $0.075 per 1M tokens. Those are not in our input/output/cached-input fields.
- Alibaba's Code Arena ranking and Work360 figures are vendor-reported. We did not verify them.
- Alibaba says open-weight, smaller-scale models from the Qwen4 series will be released later. They are not out yet.

## Alternatives to Qwen4-Max

Models from other vendors that share at least one of our groups, from each model's sourced facts (not tested by us).

- [Claude Fable 5.2](https://theagenticweekly.com/models/claude-fable-5-2) (Anthropic): Both in: Frontier models, Coding, Agents · $8 in / $40 out per 1M tokens · 1M context · released Nov 9, 2026
- [Gemini 4 Carbon](https://theagenticweekly.com/models/gemini-4-carbon) (Google): Both in: Frontier models, Agents, Coding · $2 in / $10 out per 1M tokens (introductory price until Nov 30, 2026; list $4 / $20) · released Oct 28, 2026
- [Gemini 4 Argon](https://theagenticweekly.com/models/gemini-4-argon) (Google): Both in: Frontier models, Agents, Coding · $2 in / $10 out per 1M tokens (introductory price until Nov 30, 2026; list $4 / $20) · 1.05M context · released Oct 13, 2026
- [Muse Wonder](https://theagenticweekly.com/models/muse-wonder) (Meta): Both in: Frontier models, Agents, Coding · $4 in / $20 out per 1M tokens · 1.05M context · released Oct 12, 2026
- [Muse Wonder Full](https://theagenticweekly.com/models/muse-wonder-full) (Meta): Both in: Frontier models, Agents, Coding · released Oct 12, 2026
- [GPT-6.1 Sol](https://theagenticweekly.com/models/gpt-6-1-sol) (OpenAI): Both in: Frontier models, Coding, Agents · $2 in / $10 out per 1M tokens · 1.05M context · released Sep 29, 2026
- [Best AI model for agents (Nov 2026): Claude Opus 5.5 ties for the lead on Terminal-Bench 4.1](https://theagenticweekly.com/best/ai-model-for-agents)
- [Longest context window LLMs](https://theagenticweekly.com/best/long-context-llms)

## Qwen4-Max FAQ

**When was Qwen4-Max released?**

Alibaba released Qwen4-Max on Nov 16, 2026. Source: https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783

**How much does Qwen4-Max cost?**

The listed API price is $1.5 in / $4.5 out per 1M tokens. Prices change; the source shows the current figure. Source: https://www.qwencloud.com/models/qwen4-max

**How large is the Qwen4-Max context window?**

1,000,000 tokens, with up to 393,000 output tokens. Source: https://www.qwencloud.com/models/qwen4-max

**Is Qwen4-Max open weights?**

No. The weights are not published; the licence is listed as Proprietary (API). Source: https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783

**What is the API model id for Qwen4-Max?**

qwen4-max. Source: https://www.qwencloud.com/models/qwen4-max

**Has Agentic Weekly tested Qwen4-Max?**

Not yet. Every fact on this page is Reported: read from the linked vendor pages, not measured by us.

## Sources

- [Alibaba Cloud Community: Alibaba Unveils Qwen4-Max (Nov 16, 2026)](https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783)
- [QwenCloud model page: Qwen4-Max](https://www.qwencloud.com/models/qwen4-max)
- [QwenCloud changelog: model releases](https://docs.qwencloud.com/changelog/models)
- [Alibaba Cloud Model Studio: newly released models](https://www.alibabacloud.com/help/en/model-studio/newly-released-models)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.
